摘要
为解决设计端文化逻辑缺失的问题,通过构建漳州木版年画设计要素之间的关系,探寻智能系统协同生成符合漳州木版年画文化逻辑和风格的产品新范式。首先,利用大模型构建漳州木版年画设计要素的感知联想,将漳州木版年画从四维度展开解构从而搭建源文本联想坐标系。其次,以数量化理论Ⅰ为基础,利用多元回归分析方法构建漳州木版年画感性词对与设计要素之间的决策模型,指导人工智能(AI)生成具有漳州木版年画文化逻辑和风格的新产品。结果表明:在此决策模型的指导下,能够确保生成内容符合漳州木版年画的文化设计逻辑,避免AI生成随机性,为生成式人工智能驱动的高效、精准设计提供逻辑保真约束。
To address the deficiency of cultural logic in design practices,this study establishes the relational structure among the design elements of Zhangzhou woodblock New Year paintings and explores an intelligent co-creation paradigm capable of generating products that embody their inherent cultural logic and stylistic features.First,a large model is used to construct the perceptual association of design elements,enabling a four-dimensional deconstruction of Zhangzhou woodblock New Year paintings and the establishment of a source-text associative coordinate system.Second,based on quantification theoryⅠand multivariate regression analysis,a decision-making model is developed to map affective word pairs to corresponding design elements,thereby guiding the artificial intelligenc(AI)system to generate new products aligned with the cultural semantics and aesthetic characteristics of Zhangzhou woodblock New Year paintings.Experimental results show that the proposed model effectively constrains the generative process,ensuring that the produced content adheres to the intrinsic cultural design logic while reducing randomness.This study provides a logic-preserving framework for achieving efficient and precise design generation driven by generated AI.
作者
吴晶晶
王一婷
蒋文贤
WU Jingjing;WANG Yiting;JIANG Wenxian(Xiamen Academy of Arts and Design,Fuzhou University,Xiamen 361024,China;College of Computer Science and Technology,Huaqiao University,Xiamen 361021,China)
出处
《华侨大学学报(自然科学版)》
2026年第2期247-256,共10页
Journal of Huaqiao University(Natural Science)
基金
福建省社会科学基金基础研究项目(FJ2022B104)
福建省专业学位研究生优秀教学案例(专业学位研究生教学案例库项目)(00489054)。
关键词
生成式人工智能
漳州木版年画
数量化理论Ⅰ
感知联想
风格保真约束
智能生成优化
generative artificial intelligence
Zhangzhou woodblock New Year paintings
quantitative theoryⅠ
perceptual association
style fidelity constraints
intelligent generation optimization
